1. Field of the Invention
The present invention generally relates to a technology for detecting either one of a peak level and a bottom level of an input signal, with a capability of suppressing a leak current without losing an instantaneous response characteristic.
2. Description of the Related Art
In recent years, the establishment of broadband optical access networks, which have a high speed and a high capacity, has rapidly developed to cater to the increase in data traffic due to wide use of the Internet or the like. A specific example of a high-speed optical access system is a Gigabit-Passive Optical Network (G-PON), which is able to realize transmission at a speed as high as 1 Gbps at maximum.
In an optical access system like a G-PON, an optical splitter that splits an optical fiber is provided between a plurality of apparatuses on the subscribers' side (Optical Network Units (ONUs)) and an apparatus on the station side (an Optical Line Terminal (OLT)). Thus, the one optical fiber is shared by the plurality of ONUs. In the upstream transmission from the ONUs to the OLT, the Time Division Multiple Access (TDMA) method is used. The transmission timing is controlled so that optical packets transmitted from the plurality of ONUs do not collide with one another. For this reason, it is necessary for the OLT to have an optical burst receiver that instantly identifies optical packet signals that arrive from the ONUs in a burst manner.
The strength of the optical packet signals varies highly depending on the difference in transmission loss among the ONUs. For example, the variation ranges from 100 times to 1000 times. Thus, the OLT is required to have a very wide dynamic range. Further, a preamble that is positioned at the head of an optical packet signal and that corresponds to a rise time is extremely short, for example, tens of bits. Thus, it is necessary for the OLT to receive signals of which the levels vary so highly during the course of time and to decide the logic of the signals within the preamble period. Furthermore, a guard time that is provided between optical packet signals and that corresponds to a non-transmission period is also extremely short. During the guard time, an initialization process is performed with a reset signal, and the next optical packet signal is awaited.
As explained above, it is necessary for the OLT in the optical access system to identify the optical packet signals instantly. Thus, as disclosed in Japanese Patent Application Laid-open No. 9-270755, sometimes a method is used in which the peak level and the bottom level of a signal are detected so that a level at the center between the peak level and the bottom level is used as a threshold value to decide the logic of the signals thereafter. In other words, by determining the threshold level based on the peak level and the bottom level that are detected in the preamble portion of the signal, it is possible to accurately decide the logic with regard to the data portion of the signal.
From the point of view of reducing cost, it is desirable to manufacture optical receivers that are to be used in optical access systems, based on the technology related to Complementary Metal Oxide Semiconductors (CMOS) that include a Field Effect Transistor (FET). However, in a circuit in which a high-speed CMOS is used, a problem arises where a leak current occurs due to process variations, during the detection process of the peak level and the bottom level. Particularly, when signals having the same logic in the data portion appear in succession, there is a variation in the storing of the peak level or the bottom level of the signals. Thus, the variation in the threshold value due to a leak current is noticeable when the same logics appear in succession.
One of the ways to inhibit the occurrence of the leak current is to make the gate length of the FET included in the CMOS longer. However, when the gate length is longer, the rising and the falling at the head of an optical packet signal become slower, and thus, the instantaneous response characteristic is degraded. When the instantaneous response characteristic is degraded, the determination process of a threshold level cannot be completed within the preamble portion of the signal. Accordingly, it may not be possible to make decision on the logic in the data portion accurately.